A 1-km resolution global ocean simulation promises to unveil oceanic multi-scale dynamics and climate impacts
As a crucial component of the Earth system, the ocean significantly impacts the climate due to its vast heat capacity, intricate multi-scale circulation, and considerable carbon storage capability. The ocean general circulation model (OGCM) is a numerical tool designed to solve the governing equations of oceanic fluid and thermal dynamics. It can simulate oceanic circulations and physical states, facilitating marine environmental forecasts and climate projections.
The first OGCM was developed by Kirk Bryan in 1969, featuring a resolution of several hundred kilometers. Since the onset of the 21st century, advancements in numerical methods, high-performance computing, and physical oceanography have significantly enhanced ocean modeling. Consequently, the model has become increasingly complex, with more sophisticated dynamical cores and physical parameterizations, and additional ocean-related processes such as air-sea interactions, sea ice dynamics, and tidal effects.
The challenge of high and ultra-high resolution
The horizontal resolution is essential for representing the developmental level of an OGCM, and the pursuit of higher resolution has been a persistent topic throughout the history of ocean modeling.
Conventionally, in the context of long-term global climate simulations, an OGCM is considered "high resolution" when its horizontal resolution reaches 10 km (0.1°). The primary motivation for achieving high resolution is to accurately capture oceanic mesoscale processes, which typically have spatial scales ranging from O(10 km) to O(100 km). These processes encompass the majority of the oceanic kinetic energy and substantially affect the intrinsic ocean dynamics and air-sea interactions, thereby essentially contributing to the Earth’s climate variability. High-resolution OGCMs can capture most mesoscale activities in the open ocean at low- and mid-latitudes, leading researchers to refer to them as "eddy-rich" models. However, the oceanic submesoscale, characterized by spatial scales ranging from O(100 m) to O(10 km), can closely interact with the mesoscale, thereby playing a substantial role in the climate system.3 To accurately describe the mesoscale dynamics, an ocean model must incorporate the effects of the submesoscale on the mesoscale. This presents a challenge that eddy-rich models often struggle to address.
Over the past decade, ocean models with kilometer-scale horizontal resolution have been burgeoning due to advancements in high-performance computing technology. To be distinguished from the traditional concept of "high-resolution" models, those with global average horizontal resolutions of less than 5 km are now classified as "ultra-high-resolution" models. They can fully cover the oceanic mesoscale range and capture a significant portion of oceanic submesoscale phenomena, which can closely interact with mesoscale processes through the inverse energy cascade, thereby playing a substantial role in the climate system. Ultra-high resolution OGCMs can reduce the uncertainty associated with parameterization schemes and enhance simulation fidelity, enabling more detailed studies of local processes (e.g., eddy and wavy motions) and their multi-scale interactions. These insights can guide the configuration of eddy-rich models, resulting in improved climate projections and marine environmental forecasts, particularly for high-impact, low-probability extreme events.
However, developing ultra-high-resolution models is highly challenging. The need for massive computing and storage resources makes it essential for the model to be ported onto high-performance computers with heterogeneous architectures. Developing efficient and stable parallel communication and I/O technologies, adopting customized acceleration solutions, and achieving performance portability and scalability are critical for advancing these models.
